Ted Lieu's new bipartisan AI bill targets deepfakes, AI-enabled fraud, and whistleblower protections without wading into the most divisive fight over state regulation.
Representative Ted Lieu is trying to move AI policy through Congress by focusing on the parts lawmakers can still agree on. The California Democrat, who helped lead the House Task Force on AI, introduced the American Leadership in AI Act with Republican Representative Jay Obernolte, a broad package built from more than 20 bipartisan proposals drawn from the task force's recommendations.
The bill takes aim at harmful deepfakes, AI-assisted fraud, and gaps in protection for workers who report serious AI risks. According to CNBC, the proposal is designed to avoid the fights most likely to derail it, including federal preemption of state AI laws and mandatory testing rules for advanced models. That makes the bill less sweeping than some advocates want, but potentially more realistic in a divided Congress.
Deepfakes sit at the center of that calculation. The measure would strengthen legal remedies for victims and increase penalties tied to AI-enabled fraud and impersonation, building on earlier state and federal efforts. California has already moved against non-consensual intimate images and deceptive election content, while the federal TAKE IT DOWN Act requires covered platforms to remove non-consensual intimate imagery, including AI-generated material, after notice from a victim.
The whistleblower language is just as important for AI companies. The bill would support protections for people who expose AI-related risks, misconduct, or misleading claims, an area where regulators often see problems only after employees, researchers, or contractors come forward. California's SB 53 has already pushed frontier developers toward incident reporting and internal accountability, and a federal version would raise the stakes for companies operating across state lines.
The timing is politically awkward. States have been writing their own AI rules while Washington has struggled to pass a broader framework, and the White House has pressed for fewer state-level restrictions. Lieu's bill offers a narrower route: give Congress a way to act on deepfakes, fraud, standards, research, and whistleblower protection without settling every dispute over who should regulate AI.
That restraint is also the bill's selling point. Lieu has framed the package as a consensus measure, not a full rewrite of AI oversight. It leans on proposals with bipartisan roots, including work from the House task force and earlier bills aimed at deepfake abuse, fraud, and public-sector AI readiness.
The practical effect is still significant. If the bill advances, companies building moderation tools, identity products, media platforms, and frontier models would face a clearer expectation to detect abusive synthetic media, respond to complaints, document risks, and create channels for protected reporting. The cost would not fall only on Big Tech. Startups that touch user-generated content, enterprise AI, or digital identity would need to treat compliance as part of the product, not a legal task saved for later.
Compliance Burden for AI Firms
For founders, the message is straightforward: AI governance is moving from policy debate to operating requirement. Platforms may need stronger systems for detecting, labeling, and removing harmful deepfakes. Developers may need better documentation, incident-response processes, and internal escalation paths when employees spot safety problems.
Federal rules could also reduce some of the current uncertainty. A startup serving customers in California, New Hampshire, and other active states is already navigating a patchwork of deepfake, election, and safety laws. A national framework could make that easier to manage, but only if Congress avoids wiping out stronger state protections in the process.
For enterprises, the bill reinforces a larger shift in the market. Deepfakes are no longer a theoretical risk for elections, executives, or consumers, and whistleblowers are becoming part of the accountability system around powerful AI models. Companies that build detection, provenance, audit, and reporting tools now have a clearer opening.
The bill's path is still uncertain, but its direction is not. Congress is looking for AI rules it can pass, and deepfake abuse gives lawmakers a problem that is visible, personal, and difficult to ignore. If Lieu and Obernolte can keep the coalition together, the first durable federal AI framework may arrive through targeted safeguards rather than one grand regulatory bargain.
Also read: Microsoft drops OpenAI exclusivity and ends revenue share in landmark deal restructure • Cursor's Claude agent wipes production database and backups in 9 seconds • Google employees demand AI red lines as Pentagon contracts reach billions
